Detection and Identification of Gross Errors in Data Obtained in the Large-Scale Process Utilizing Mass Balances

Abstract

It is shown that the most powerful equation (the most powerful cocycle) for detecting gross errors can be easily obtained if all the measured variables in the process are constrained by only mass balances. An algorithm is developed for the rational decomposition of the problem of detection and identification of gross errors in the large-scale process and the efficiency of this algorithm is demonstrated by applying it to numerical experiments with the use of a simplified refinery model with 22 flowmeters.